Abstract
There has been an increasing trend to utilize short cognitive batteries for the diagnosis of dementia. Most of these batteries have been designed in countries with high standards of education and are less suitable for populations with low levels of education. We developed a battery that has been previously shown to be highly accurate in the diagnosis of dementia in individuals with low levels of education. The accuracy of this battery for patients with higher levels of education is unknown.
Objectives:
To evaluate the accuracy of a brief cognitive battery in the diagnosis of Alzheimer's disease (AD) in subjects with medium and high levels of schooling, and to develop a mathematical model that includes the most discriminative tests.
Methods:
Seventy-three mildly demented patients with probable AD and 94 control subjects were evaluated. Sixty patients and 60 controls were randomly selected to generate a mathematical model including the most discriminative tests of the battery using logistic regression. The model was back-tested for the remaining sample of patients and controls.
Results:
Delayed recall, learning and category fluency tests were included in a mathematical model that obtained an area of 0.917 in the ROC curve in the back-testing. Inter-rater reliabilities of these tests were high (kappa>0.8).
Conclusions:
This model showed a high accuracy for the diagnosis of mild AD in patients with medium and high educational levels. Future studies with more heterogeneously educated individuals are necessary to investigate whether the educational level (number of years at school) should also be included in the model.
Key words:
Alzheimers disease; dementia; diagnosis; education; brief cognitive battery; neuropsychological tests.